View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by King's Research Portal

King’s Research Portal

DOI: 10.1186/s12905-019-0744-z

Document Version Publisher's PDF, also known as Version of record

Link to publication record in King's Research Portal

Citation for published version (APA): Chomat, A. M., Andersson, N., Menchu, A. I., Ramirez-Sea, M., Pedersen, D., Bleile, A., ... Araya Baltra, R. (2019). Women’s circles as a culturally safe psychosocial intervention in Guatemalan indigenous communities: a community-led pilot randomised trial. BMC Women's Health, 19(1), [53]. https://doi.org/10.1186/s12905-019- 0744-z

Citing this paper Please note that where the full-text provided on King's Research Portal is the Author Accepted Manuscript or Post-Print version this may differ from the final Published version. If citing, it is advised that you check and use the publisher's definitive version for pagination, volume/issue, and date of publication details. And where the final published version is provided on the Research Portal, if citing you are again advised to check the publisher's website for any subsequent corrections. General rights Copyright and moral rights for the publications made accessible in the Research Portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognize and abide by the legal requirements associated with these rights.

•Users may download and print one copy of any publication from the Research Portal for the purpose of private study or research. •You may not further distribute the material or use it for any profit-making activity or commercial gain •You may freely distribute the URL identifying the publication in the Research Portal Take down policy If you believe that this document breaches copyright please contact [email protected] providing details, and we will remove access to the work immediately and investigate your claim.

Download date: 09. May. 2019 Chomat et al. BMC Women's Health (2019) 19:53 https://doi.org/10.1186/s12905-019-0744-z

RESEARCHARTICLE Open Access Women’s circles as a culturally safe psychosocial intervention in Guatemalan indigenous communities: a community-led pilot randomised trial Anne Marie Chomat1,2* , Aura Isabel Menchú2, Neil Andersson1,3, Manuel Ramirez-Zea4, Duncan Pedersen5ˆ, Alexandra Bleile6, Paola Letona4 and Ricardo Araya6

Abstract Background: Indigenous Maya women in Guatemala show some of the worst maternal health indicators worldwide. Our objective was to test acceptability, feasibility and impact of a co-designed group psychosocial intervention (Women’s Circles) in a population with significant need but no access to mental health services. Methods: A parallel group pilot randomised study was undertaken in five rural Mam and three periurban K’iche’ communities. Participants included 84 women (12 per community, in seven of the communities) randomly allocated to intervention and 71 to control groups; all were pregnant and/or within 2 years postpartum. The intervention consisted of 10 sessions co-designed with and facilitated by 16 circle leaders. Main outcome measures were: maternal psychosocial distress (HSCL-25), wellbeing (MHC-SF), self-efficacy and engagement in early infant stimulation activities. In-depth interviews also assessed acceptability and feasibility. Results: The intervention proved feasible and well accepted by circle leaders and participating women. 1-month post- intervention, wellbeing scores (p-value 0.008) and self-care self-efficacy (0.049) scores were higher among intervention compared to control women. Those women who attended more sessions had higher wellbeing (0.007), self-care and infant-care self-efficacy (0.014 and 0.043, respectively), and early infant stimulation (0.019) scores. Conclusions: The pilot demonstrated acceptability, feasibility and potential efficacy to justify a future definitive randomised controlled trial. Co-designed women’s groups provide a safe space where indigenous women can collectively improve their functioning and wellbeing. Trial registration: ISRCTN13964819. Registered 26 June 2018, retrospectively registered. Keywords: Maternal mental health, Indigenous women, Guatemala, Participatory research, Women’scircles,Co-design, Cultural safety

Background emotional, and social development [1–4]. Maternal mental Perinatal mental disorders – depression, anxiety and som- health has been linked to reduced responsiveness in care- atic disorders – can be detrimental to women’shealth, giving and higher rates of behavioral problems in children pregnancy outcomes and infant neurological, cognitive, [5] and young adults [6]. Maternal anxiety – which indi- genous women may be at increased risk for [7] – has been * Correspondence: [email protected]; [email protected] associated with preterm birth [8] and, in low-income set- ˆ Postmortem tings, maternal depression has been associated with low 1Participatory Research at McGill (PRAM), Department of Family Medicine, McGill University, 5858 Chemin de la Côte-des-Neiges-3rd floor, Suite 300, birth weight, childhood stunting, higher rates of diarrheal Montréal, QC H3S 1Z1, Canada diseases and poor cognitive development in young chil- 2 CIET International Guatemala, 5ª calle 14-35, apartamento 304, Edificio Las dren [1, 9]. Tapias, zona 3, Quetzaltenango, Guatemala Full list of author information is available at the end of the article

© The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Chomat et al. BMC Women's Health (2019) 19:53 Page 2 of 15

A systematic review reported perinatal mental disorders requested a group intervention – Women’s Circles – were common in low- and lower-middle-income countries that could help and provide support for women in their (LMIC), affecting 16% of pregnant women and 20% of communities, following earlier involvement in a partici- women in the postpartum period [10]. Indigenous women patory research project with the lead author of this experience higher rates of partner abuse than paper [17]. In each community, local leaders steered non-indigenous women, for whom partner violence is an group processes. We chose a participatory research ap- especially strong predictor of poor mental health [11]. proach [33, 34] to optimize community engagement and Protective factors include relative social and economic ad- optimize cultural safety, acceptability and feasibility. vantage, formal education, secure employment, reproduct- ive health services, belonging to the ethnic majority, and Trial design having a respectful, trustworthy intimate partner [10]. The design was a parallel group pilot randomised study. Women may also be better able to counter stress if they have high self-esteem and self-efficacy [12], effective social Setting support [13], and an ability to problem-solve [8]. Five rural Mam communities in San Juan Ostuncalco Guatemala’s indigenous women manifest some of the municipality (25 km from Quetzaltenango city; popula- worst health indicators worldwide [14]; three in four live tion 1000-4000) and three periurban K’iche’ communi- below the poverty line [15]. Women of childbearing age ties in Quetzaltenango city (population 4000-16,000) living in indigenous areas show the highest rates of de- were selected as study sites, based on prior collabora- pression and anxiety in the country [16]. In rural indigen- tions with the first author and local women leaders’ ous Mam communities in the Western Highlands, lower expressed interest in participating. household wealth, psychological distress, ineffective social support, inequality in decision-making, and experience of violence were consistent determinants of maternal stress, Co-design assessed via salivary cortisol, and infant stunting in the Ten six-hour workshops scheduled monthly with 16 cir- first 6 months of life [17]. Guatemala’s national health sys- cle leaders defined the transdiagnostic (addresses a range tem provides limited access to mental health services; of mental health issues) intervention. Circle leaders col- there are no formal mental health promotion and preven- lectively chose a project name and logo; developed a the- tion programs, and limited involvement of service users ory of change; mapped community needs, resources, and and families in mental health systems [18, 19]. The Guate- stakeholders; and pilot tested group methodologies. malan civil war and long history of racial discrimination Group activities drew on games (dinámicas), art-based places indigenous populations at an additional disadvan- methods (drawing, role play, music) and group psycho- tage in terms of access to health services [20, 21]. social therapy (active listening, emotion management, Recent research has demonstrated the feasibility of breathing and relaxation exercises, problem solving, psychosocial interventions for perinatal mental health in popular education) to build trust, self-esteem, and social ’ non-specialized health-care settings using psychoeduca- cohesion. Women s interest in developing tion [5, 22–24] cognitive restructuring [25, 26], livelihood-sustaining skills prompted us to also incorpor- problem-solving [25, 27, 28], behaviour activation [27], ate productive activities (i.e. doll-making, crochet, cook- activating social networks [28, 29], and skilled parenting ing) as vocational therapy and potential income practices [30–32]. Few of these interventions have been generation. tested in Latin America [27, 29] and none in indigenous populations. We addressed this knowledge gap for a Intervention population at special disadvantage of maternal mental Additional file 1: Table S1 outlines the contents of the health disorders through the co-design of a culturally 10 sessions that followed a standard format. Pre-sessions safe perinatal group psychosocial intervention compat- involved toy-making of dolls, books or rattles mothers ible with indigenous traditions – Women’s Circles. could use to stimulate and play with their infants. Ses- The objective of this pilot randomised study was to as- sions started with an inclusive participant-led prayer, sess co-designed Women’s Circles’ in terms of accept- followed by a prior session recap. A group game or ability, feasibility and proof-of-concept in preparation dinámica served as an icebreaker. Activities that enabled for a future definitive trial. personal and group reflection (drawing, dramatization) led to sharing lessons-learned, aspirations and personal Methods experiences. A closing dinámica released tensions or Community involvement promoted relaxation, through guided meditation or deep Local women in the Mam communities (community breathing exercises. Sessions concluded with a collective health workers and traditional midwives, or comadronas) embrace. Sessions took place every fortnight in settings Chomat et al. BMC Women's Health (2019) 19:53 Page 3 of 15

of participant’s choosing (i.e. house, community center), and their own familiarity with local women. This and lasted on average 2 h. method seemed realistic and feasible for future imple- The intervention extended over 5 months, with ses- mentation of the study. sions taking place every other week. Surveys Control All participants underwent baseline and follow-up surveys. Control women did not receive an intervention but were At enrolment, eligible women gave informed consent, a invited to join a Women’s Circle when the trained female interviewer (fluent in Spanish and Mam in post-intervention assessment was complete. Mam communities) administered the questionnaire, and nutritionists measured height and weight of her youngest Circle leaders child. A follow-up home-based assessment used the same The 16 circle leaders were identified based on prior col- questionnaire 1-month post-intervention. All instruments laborations and expressed interest and invited to underwent pilot testing and semantic validation in Span- co-design and co-facilitate the intervention. Nine were ish. As few could read Mam or K’iche,’ no Maya transla- former community health workers (CHWs), six were tions were performed; instead, data collectors agreed on comadronas and one a community leader (former vocabulary to be used with non-Spanish speakers. Surveys mayor). Aged 27 to 70 (mean 47.4 ± 14.5) years, one had took between 20 and 30 min to complete. no formal schooling, six had incomplete and five com- pleted primary schooling and four had incomplete sec- In-depth interviews ondary schooling. Training by our research team lasted Post-intervention, a trained, bilingual Mam-Spanish fe- 50 h. After their own researcher-led 10-session Women’s male interviewer conducted in-depth interviews of 14 cir- Circle, where the 16 leaders acted as participants, they cle leaders and of two women participants in each of the practiced session delivery (2/week, over 5 weeks). Add- seven communities still participating in the intervention, itional training included crisis response, counselling, after obtaining informed consent. The script-based inter- group facilitation and self-care skill-building. They re- views lasted between 20 and 45 min, were conducted in a ceived per diems of 50 quetzals (seven USD per day). All location of the women’s choice, and were audio-recorded. training activities were carried out in the leaders’ homes, on a rotating basis, as per their preference. Outcomes Primary outcomes Intervention fidelity In the week preceding sessions, circle leaders joined a  Maternal symptoms of depression and anxiety over practice round. Ongoing support included phone the last month, using the Hopkins Symptom debriefing and direct observation of a random sample of Checklist-25 [35] (HSCL-25), a symptom inventory sessions, carried out with all leaders by our research composed of a 10-item anxiety cluster, a 13-item de- team. The research team and more experienced circle pression cluster, and two additional somatic symp- leaders accompanied others facing difficulties. toms. Each item scores on a scale from one (not at all) to 4 (extremely); item scores can be summed to Participant selection provide an estimate of the severity of anxiety and A checklist for participant eligibility included being depression symptomatologies. A higher score indi- pregnant or under 2 years postpartum and having at cates greater distress. least one of the following conditions: socioeconomic dis-  Maternal wellbeing, using the Mental Health advantage, domestic violence, difficult interpersonal rela- Continuum Short Form [36] (MHC-SF), comprised tionships, poor social support, or psychological distress. of 14 items representing the three dimensions of These criteria were based on known risk factors [10], wellbeing: emotional, social and psychological. Each circle leaders’ assessment of what constituted maternal item scores on a scale from zero (never) to four vulnerability, and prior research in nearby Mam commu- (always), based on experiences in the previous nities [17]. Circle leaders visited eligible participants, ex- month, allowing for continuous assessment of plained the intervention and invited their participation. positive mental health. A higher score indicates We originally intended to recruit women who scored greater wellbeing. high on an initial screening test for symptoms of depres- sion and anxiety; however, the absence of primary health Secondary outcomes care services in the target communities made it difficult to screen this population. Instead, the leaders thought it  Self-efficacy measurement used a four-item subscale preferable to select participants based on known need measuring self-efficacy in childcare (feeding, caring Chomat et al. BMC Women's Health (2019) 19:53 Page 4 of 15

and cleaning, playing and talking, helping recover rate of women participant retention, and number of ses- from illness) and a four-item subscale measuring sions attended. self-efficacy in self-care (overcoming daily problems; staying calm when worried, nervous, or afraid; find- Randomisation ing reliable people for support; dedicating time to Sequence generation and implementation of randomization herself). Each item scores on a scale from zero (I We used a non-computerised randomisation process. can’t do it) to three (I can do it), allowing for con- For each participating community, names of consenting tinuous assessment of childcare self-efficacy, self- women were put in a box and 12 names were drawn care self-efficacy, and total self-efficacy. A higher randomly to join the intervention group. Remaining score indicates greater self-efficacy. names were allocated to the control group with a de-  Mother’s engagement in early infant stimulation, layed circle intervention. using six items from the UNICEF Multiple Indicator Cluster Survey Early Child Development module Allocation concealment and masking capturing adult-child interactions [37], assessing In a study of this nature it is virtually impossible to keep whether mothers engaged with her infant in six dif- allocation to groups concealed after the intervention ferent activities (e.g., reading, singing, playing, talk- starts. However, we made no announcements as to the ing) over the preceding 3 days. Each item scored as allocation to any of the participants. Given the nature of zero (no) or one (yes). The cumulative number of the intervention, masking was not possible either. activities was used as a continuous variable for ana- lyses. Only women who had a child under 2 years Sample size old participated in this questionnaire. A higher score A total of 176 women in the eight participating commu- indicates greater involvement in early infant stimula- nities met eligibility criteria. One community and its 16 tion activities. participants withdrew from the study prior to random- isation. Without subsampling, in the remaining seven Sociodemographic information Included: date of birth, communities, we allocated 84 women to the intervention age, ethnicity (Mam, K’iche’, non-indigenous), language arm (one Women Circle of 12 women per community) proficiency (Mam, K’iche’, Spanish), marital status (mar- and the remaining 71 were allocated to a control arm ried, informal-union, single, separated, widowed), occu- (with women spread over the seven communities). pation, formal schooling (none, primary, secondary, higher), parity and access to health providers (CHWs, Data analysis doctors/nurses, traditional healers, religious leaders), rel- Data entry and security atives they lived with, and ownership of: electricity, fau- Manual, double data entry of questionnaires minimized cet, toilet, refrigerator, computer, mobile phone, errors. Quantitative analyses relied on SPSS Statistics television, motorcycle/bicycle, car/truck or separate Program (version 22.0) for all primary and secondary room for children. analyses, and on CIETmap open-source software for supplementary analyses; all followed the intention to Measures of acceptability Acceptability was assessed in treat principle. In-depth interviews were transcribed ver- post-intervention in-depth interviews when participants batim in Spanish and analysed using MAXQDA 11 (ver- were asked whether they were satisfied with the inter- sion 11.2.1). vention, would recommend it to other women, and would have preferred it to be any different. We also Baseline analysis asked participants to report on barriers to participation; Descriptive statistics of demographic and psychosocial circle leaders were asked what strategies they used to (outcome) measures were compared across groups overcome these. (intervention vs control). All comparisons are accom- panied by their corresponding p-values. Measures of feasibility Feasibility was assessed during post-intervention in-depth interviews. Circle leaders Primary analyses were asked whether they felt comfortable in their ability We used independent sample t-tests comparing inter- to lead the Women Circles, had received enough train- vention vs control on the four mean psychosocial scores ing and support, felt that implementation logistics were and their sub-scores to test for the potential effect of the appropriate (i.e. session frequency, location and length, intervention compared to the control group. All com- materials, compensation), and what they might change. parisons are accompanied by their corresponding Objective data included: rate of circle leader retention, p-values. Chomat et al. BMC Women's Health (2019) 19:53 Page 5 of 15

Secondary analyses Ethical review We compared outcome variables across groups using Research ethics boards at the Douglas Mental Health multiple linear regression analyses on each of the four University Institute (McGill University) and Institute of outcome variables (a) prior to adjusting for other vari- Nutrition of Central America and Panama (INCAP) in ables, (b) after adjusting for baseline values and (c) Guatemala approved the study. Community leaders and adjusting for baseline values, area of residence (rural vs. the Ministry of Health also granted permission. Each periurban), and maternal age; the latter data is shown. participant provided fully informed consent (signature We also explored for a potential “dose response” (associ- or thumbprint) at enrolment. Circle leaders recruited ation between total number of sessions attended and and obtained informed consent of all participants. outcome variables) among intervention women, using number of sessions attended as a continuous variable Results (zero to ten), performing the same analyses as described Participant flow is shown in Fig. 1. above. For all multiple regressions, we tested the follow- Fewer than 3% of eligible mothers did not provide ing assumptions: Absence of multicollinearity (variance consent. One periurban community and both its leaders inflation factor, VIF < 2.5); Independence of residuals and participants dropped out prior to randomisation due (Durbin Watson statistic between one and three); Vari- to local women’s time constraints related to employ- ance of residuals, or homoscedasticity (scatterplot of re- ment, resulting in a final sample size of 155 women (84 siduals); and Normal distribution of residuals (normal intervention, 71 control) in seven communities, and 14 P-P plot of residuals). Unstandardized coefficients (B) ± circle leaders. standard errors (SE) and P-value are reported. Recruitment Supplementary analyses The pilot was successfully conducted within the stipu- Given high baseline heterogeneity between rural and lated period of 5 months in both settings. Recruitment peri-urban areas, we used generalised estimating equa- strategies were successful; there were many women who tion (GEE) for Logistic Regression in the R package Zelig met eligibility criteria and were interested in participat- [38] in an exchangeable correlation structure (logit.gee ing. No major untoward or unexpected incidents were model, 1000 simulations, robust 95%CI) to evaluate for a reported. potential cluster effect on psychosocial outcome vari- ables. We created binomial scores for all primary and Numbers analysed secondary outcome variables, using mean All recruited mothers were invited to complete the sur- pre-intervention scores as cut-off (0 if ≤ mean, 1 if > vey; out of the 155 study participants, 147 (81 interven- mean). Analyses were adjusted for pre-intervention tion, 66 control) completed it at baseline, and 121 (68 scores and maternal age. Odds ratios (OR), robust 95% intervention, 53 control) post-intervention. confidence intervals (CI) and number needed to treat The 84 intervention mothers attended a mean of 4.6 ± (NNT) are reported. 3.6 sessions: 19% did not show up for any sessions, 17% attended one session, 10% attended two to four sessions Missing data and 55% five or more. We imputed missing data for individual items of the psychosocial questionnaires by calculating the mean of Baseline data the remaining questionnaire items and replacing the Study participants were easily identified based on the de- missing data with that value. All other missing data (i.e. fined eligibility criteria. The majority (73%) were selected socio-demographics) were ignored. for living in extreme poverty, 40% for experiencing psy- chosocial distress, and 3% for having family problems Qualitative analyses (categories are non-exclusive). AB and PL independently analysed the data. MAXQDA Mean maternal age was 26.2 ± 6.4 yrs. (15 to 43) 11 (version 11.2.1) was used to organize the data and (Table 1). Most (95%) rural mothers self-described as code the transcripts via thematic content analysis [39]. Mam, and 74% periurban mothers as K’iche’. The major- Codes included dimensions of acceptability (i.e. affective ity reported living in economically insecure households attitude, burden, self-efficacy, perceived effectiveness, (57%); 59% had a stunted lastborn child, and 4% a coherence), feasibility (i.e. barriers to participation) and wasted child. dimensions of expected effectiveness (i.e. self-esteem, There were no significant baseline differences between self-efficacy, social support, knowledge exchange, emo- intervention and control women, in either sociodemo- tional wellbeing). AMC reviewed both analyses and ex- graphic (Table 1) or primary or secondary outcome tracted key dimensions and quotes for publication. (Table 2) measures. Chomat et al. BMC Women's Health (2019) 19:53 Page 6 of 15

Enrollment Eligible women (n=176)

Excluded (n=21) • Declined to participate (n=5) • Community dropped out early on due to logistical difficulties in implementation of intervention (n=16)

Randomized (n=155)

Allocation Allocated to intervention (n=84) Allocated to control (n=71) • Received allocated intervention (n=84) • Received allocated intervention (n=71)

Follow-up Lost to follow-up (did not have time to attend Lost to follow-up (did not have time to complete circles or complete evaluation) (n= 16) evaluation) (n=18)

Assessment Primary analysis (n=68) Primary analysis (n=53) Secondary analysis (n=67) Secondary analysis (n=52) Supplementary analysis (n=67) Supplementary analysis (n=52)

Fig. 1 Participant flow diagram

Since we found significant differences between rural post-intervention MHC-SF score (greater wellbeing) and urban populations, we conducted additional ana- than were control women (p-value 0.011) (Table 4). lyses to understand this situation better. Rural compared They were also significantly more likely to have a higher to periurban women were more likely to have higher self-care sub-score score (greater self-efficacy in parity (2.8 ± 1.8 vs. 2.0 ± 1.2; p-value 0.008), be in an in- self-care) after controlling for baseline score only (data formal union (60 vs. 10%, p-value < 0.001) and have less not shown; p-value 0.028); however, this difference be- schooling (20 vs. 7% never attended school, p-value < came less apparent in the fully adjusted model (p-value 0.001). They were less likely to be employed (1 vs. 23%, 0.056). All assumptions are presented in Additional file 2: p-value < 0.001) and to own refrigerators, cell phones, Table S2. televisions, car or trucks, or have a separate room for Multiple linear regression analyses revealed several sig- children (p-value < 0.05). Rural women had significantly nificant associations between number of sessions lower anxiety sub-scores (15.5 ± 4.7 vs. 17.4 ± 4.8; attended and primary and secondary outcome variables p-value 0.045), total HSCL-25 scores (36.1 ± 11.4 vs. (Table 5). Having participated in a greater number of 40.1 ± 9.4; p-value 0.036), wellbeing scores (38.7 ± 12.3 sessions was associated with having (1) a higher vs. 51.5 ± 9.2; p-value < 0.001) and infant stimulation MHC-SF score (greater wellbeing; p-value 0.007), (2) a scores (1.7 ± 2.4 vs. 5.9 ± 3.5; p-value < 0.001). higher self-care sub-score (greater self-efficacy; p-value 0.014); (3) a higher infant-care sub-score (greater Outcomes and estimation self-efficacy; p-value 0.043), and (4) a higher infant Primary analyses stimulation score (greater maternal participation in early Post-intervention, intervention compared to control infant stimulation; p-value 0.019). All assumptions are women had a significantly higher MHC-SF score (greater presented in Additional file 2: Table S2. wellbeing; 45.8 ± 10.5 vs. 40.2 ± 12.5; p-value 0.008) and a significantly higher self-care sub-score (greater self-efficacy Supplementary analyses in self-care; 9.2 ± 2.5 vs. 8.4 ± 2.0; p-value 0.049) (Table 3). GEE analyses revealed several significant associations be- However,therewerenodifferencesintheHSCL-25scores tween study arm allocation and primary and secondary or sub-scores, in the total self-efficacy score, or in the en- outcome variables post-intervention, clustering by site gagement in infant stimulation activities score. (rural versus periurban) and adjusting for pre-intervention score and maternal age (Table 6). Rela- Secondary analyses tive to control, the intervention increased both the Multiple linear regression analysis revealed that inter- MCH-SF score (greater wellbeing, OR 2.01, 95% CI vention women were more likely to have a higher 1.39–2.89) and the self-care self-efficacy sub-score Chomat et al. BMC Women's Health (2019) 19:53 Page 7 of 15

Table 1 Baseline characteristics of participants in intervention vs. control groups: Mean, standard deviation and sample size (N) for continuous variables, and percent and sample size (N) for categorical variables Control Intervention P-value a Maternal age, yrs 26.2 ± 6.5 (65/66) 26.2 ± 6.3 (79/81) 0.953 Parity, # 2.6 ± 1.9 (62/66) 2.5 ± 1.6 (80/81) 0.802 Ethnicity, self-reported Not indigenous 7.3% (4/55) 12.7% (9/71) 0.327 Mam 78.2% (43/55) 66.2% (47/71) K’iche’ 14.5% (8/55) 21.1% (15/71) Reproductive status Infant 0 to 2 years old 75.8% (50/66) 82.7% (67/81) 0.298 Pregnant 32.3% (21/65) 19.8% (16/81) 0.083 Marital status Married 47.0% (31/66) 39.5% (32/81) 0.218 Informal union, living with partner 47.0% (31/66) 45.7% (37/81) Single/widowed 6.1% (4/66) 14.8% (12/81) Formal schooling None 20.0% (11/55) 14.1% (10/71) 0.921 Incomplete primary 34.5% (19/55) 42.3% (30/71) Complete primary 23.6% (13/55) 21.1% (15/71) Incomplete secondary 14.5% (8/55) 11.3% (8/71) Complete secondary 5.5% (3/55) 9.9% (7/71) Higher education 1.8% (1/55) 1.4% (1/71) Profession Housewife 92.4% (61/66) 95.1% (77/81) 0.732 Living with... Mother 18.2% (12/66) 23.5% (19/81) 0.436 Mother-in-law 33.3% (22/66) 32.1% (26/81) 0.874 Partner 84.8% (56/66) 80.2% (65/81) 0.467 Economic security Economically insecure household 57.4% (31/54) 55.7% (39/70) 0.850 Household assets Electricity 7.3% (4/55) 14.1% (10/71) 0.228 Refrigerator 20.0% (11/55) 26.8% (19/71) 0.377 Computer 5.5% (3/55) 4.2% (3/71) 1.000 Cellphone 12.7% (7/55) 16.9% (12/71) 0.619 TV 58.2% (32/55) 57.7% (41/71) 0.961 Separate room for children 32.7% (18/55) 28.2% (20/71) 0.580 Motorcycle or bicycle 47.8% (11/55) 52.2% (12/71) 0.655 Car or truck 25.5% (14/55) 46.5% (33/71) 0.016 Toilet 96.4% (53/55) 98.6% (70/71) 0.580 Faucet 87.3% (48/55) 87.3% (62/71) 0.993 Access to physical/emotional health provider No one 18.4% (9/49) 34.8% (23/66) 0.051 Health worker 46.9% (23/49) 45.5% (30/66) 0.245 Doctor or nurse 63.3% (31/49) 45.5% (30/66) 0.425 Traditional healer 8.2% (4/49) 12.1% (8/66) 0.265 Religious leaders 20.4% (10/49) 19.7% (13/66) 0.595 Family 73.5% (36/49) 56.1% (37/66) 0.740 Infant nutritional status Chomat et al. BMC Women's Health (2019) 19:53 Page 8 of 15

Table 1 Baseline characteristics of participants in intervention vs. control groups: Mean, standard deviation and sample size (N) for continuous variables, and percent and sample size (N) for categorical variables (Continued) Control Intervention P-value a Stunting 61.2% (30/49) 56.7% (38/67) 0.349 Wasting 8.2% (4/49) 1.5% (1/67) 0.161 aP-value adjusted for multiple comparisons, significance level set at p < 0.001

(greater self-efficacy, OR 2.02, 95% CI 1.22–3.35); and other projects) obligations. The majority were satisfied with decreased the HSCL-25 score (lessened psychosocial dis- the training and supervision received, feeling it strength- tress, OR 0.86, 95% CI 0.85–0.86). Numbers needed to ened their knowledge and leadership skills and adequately treat (NNT) were 6, 6 ad 33, respectively. prepared them. The manual was a useful reference. They reported initiating various engagement strategies, including: Acceptability and feasibility visiting mothers in their homes; arranging the timing, dur- Circle leaders as delivery-agents ation and location of the sessions at the mothers’ conveni- All mothers felt comfortable with the circle leaders and ence; preparing food; and adapting and creating new that they could trust them. activities to meet mothers’ needs and interests (including A few circle leaders had initially been hesitant about their between sessions with productive activities). ability to lead a group intervention. Post-intervention, all expressed satisfaction from their role and saw it as a posi- Views about the intervention tive experience. They appreciated learning new knowledge All participants and leaders thought the intervention and skills, helping other women, and making a meaningful was a positive experience, and most requested it be con- contribution to their community. One young leader tinued. One woman shared, “We learned many new expressed, “I am happy, because now I am no longer afraid things from one another and we also had fun and [of leading the sessions]. Initially I was very anxious, but laughed with other women and shared what we felt in after a while it became easier, and the women liked all the our hearts. Coming to the Circles helped us forget our sessions, and some of them now come to me to talk about worries for a while and spend a pleasant moment.” their problems”. The majority of women appreciated the manner in Most leaders reported having time to fulfill their role. Oc- which the sessions were held, including the play- and casional scheduling difficulties were related to personal (i.e. arts-based activities, which enabled them to “relax, re- religious) or work (i.e. agriculture, attending deliveries, lease tensions and feel positive emotions”. Some themes (i.e. early infant stimulation) and activities (i.e. cooking, Table 2 Baseline characteristics: T-test comparison of handicraft-making) were especially appreciated. Several psychosocial scores of women participants, in intervention vs. reported sharing content with trusted family, most often control groups: Mean, standard deviation and sample size (N) that touching on self-esteem, childrearing, early infant Psychosocial scores a Psychosocial scores stimulation and the family economy: “When [this par- p Control Intervention -value ticipant] returns home, her mother-in-law asks her what Psychosocial distress score (HSCL-25) she learned, and she shares what she learns with her, Anxiety sub-score 15.4 ± 4.5 (66) 16.5 ± 5.0 (81) 0.144 and she shares the exercises with her” [leader]. Depression sub-score 20.1 ± 6.7 (66) 22.1 ± 6.9 (81) 0.083 Only in one periurban community did women oppose Total score 35.5 ± 10.5 (66) 38.6 ± 11.3 (81) 0.088 the sessions on inter-partner violence and reproductive health. Wellbeing score (MHC-SF) The leaders reported that session delivery took longer Total score 41.7 ± 12.8 (66) 42.6 ± 12.8 (81) 0.661 than anticipated and suggested reducing content or split- Self-efficacy score ting them into various sessions. Most mothers and Infant care sub-score 11.0 ± 1.6 (49) 10.6 ± 2.1 (66) 0.271 leaders suggested holding more frequent productive Self-care sub-score 8.0 ± 2.3 (64) 8.0 ± 2.3 (80) 0.890 workshops (i.e. every other week), alternating them with Total score 18.9 ± 3.2 (48) 18.4 ± 3.9 (66) 0.488 the more theoretical ones. Infant stimulation engagement Total score 2.8 ± 3.6 (49) 2.7 ± 2.9 (66) 0.818 Effectiveness aA higher psychosocial distress score (HSCL-25) indicates greater distress; a higher wellbeing score (MHC-SF) indicates greater wellbeing; a higher auto- Self-esteem and agency Most participants and leaders efficacy score indicates greater self-efficacy; a higher Infant stimulation engagement score indicates greater maternal engagement in infant felt the Circles had positively impacted their self-esteem. stimulation activities Many women said they learned to value themselves (“As Chomat et al. BMC Women's Health (2019) 19:53 Page 9 of 15

Table 3 Primary analysis: T-test comparison of post-intervention Table 5 Secondary analysis: Multiple linear regression models of psychosocial scores of women participants, in intervention vs. number of sessions attended (0 to 10) and the psychosocial control groups: Mean, standard deviation and sample size (N) health scores, among mothers in the intervention arm, adjusted Psychosocial scores a Post-intervention psychosocial scores for maternal age, area of residence and baseline score: B = unstandardized coefficient, SE = standard error and sample Control Intervention p-value size (N) Psychosocial distress score (HSCL-25) Psychosocial scores a B ± SE (N) P-value Anxiety sub-score 15.3 ± 4.7 (53) 15.8 ± 4.4 (68) 0.561 Psychosocial distress score (HSCL-25) b 0.225 ± 0.242 (66) 0.358 Depression sub-score 20.4 ± 7.0 (53) 21.0 ± 6.6 (68) 0.658 Wellbeing score (MHC-SF) 0.819 ± 0.294 (66) 0.007 Total score 35.7 ± 11.4 (53) 36.7 ± 10.7 (68) 0.608 Self-efficacy score: Self-care sub-score 0.202 ± 0.080 (66) 0.014 Wellbeing score (MHC-SF) Self-efficacy score: Infant care sub-score 0.141 ± 0.067 (47) 0.043 Total score 40.2 ± 12.5 (54) 45.8 ± 10.5 (68) 0.008 Infant stimulation score 0.165 ± 0.068 (46) 0.019 Self-efficacy score aIncreases in the HSCL-25, MHC-SF, self-efficacy and infant stimulation scores Infant care sub-score 10.9 ± 1.6 (38) 11.2 ± 1.7 (59) 0.446 indicate greater distress, greater wellbeing, greater self-efficacy, and greater maternal engagement in infant stimulation activities, respectively. b Self-care sub-score 8.4 ± 2.0 (52) 9.2 ± 2.5 (68) 0.049 Depression and anxiety sub-scores with similar findings, namely non- Total score 19.4 ± 3.2 (38) 20.5 ± 3.7 (59) 0.130 significant association with attendance; data not shown Infant stimulation engagement Another woman explained: “A woman is afraid of her Total score 1.4 ± 2.0 (37) 1.9 ± 2.0 (59) 0.241 husband, he is in charge. But I learned a woman also has aA higher psychosocial distress score (HSCL-25) indicates greater distress; a the right to speak or express her opinions. This is what I higher wellbeing score (MHC-SF) indicates greater wellbeing; a higher auto- am happiest about, because a woman may be pretty or efficacy score indicates greater self-efficacy; a higher Infant stimulation engagement score indicates greater maternal engagement in infant ugly or whatever, she has the right to speak her mind stimulation activities too.” A leader described how one participant, who used to have low self-esteem, told her husband that she would women, we hardly ever value ourselves, here in our decide what to do with her life, and not just follow his community”), as captured by one participant: or his mother’s wishes: “She had the courage to speak up, and she decided where she would deliver her baby.” I used to place much importance on what others said One leader mentioned that her participation had about me, and this made me feel bad or sad. My “helped her become stronger, braver, and to not let my- mother-in-law often told me I was stupid, not worth self be overcome by anything.” Another shared, anything. Now I take a bit of time every day to see myself and make myself feel better about who I am The activity I prefer is when women draw their and what I know how to do. I now try not to place so personal map, in the shape of a tree. We are as plants. much importance on what they say and excuse myself We have roots too. It is only that sometimes we don’t from people who are being offensive. I go for a walk value who we are, what we hold in our arms, what we when they start insulting me at home.

Table 6 Supplementary analysis: General Estimating Equation for Logistic Regression of study arm allocation (intervention vs. Table 4 Secondary analysis: Multiple linear regression models of control) and psychosocial health scores, clustering for area of study arm allocation (intervention vs. control) and psychosocial residence (rural vs. periurban) and adjusted for baseline score health scores, adjusted for maternal age, area of residence and and maternal age: OR = odds ratio, NNT = number needed to baseline score: B = unstandardized coefficient, SE = standard treat, CI = confidence interval error and sample size (N) Psychosocial scores a OR (95% CI) NNT (95% CI) Psychosocial scores a B ±SE P-value Psychosocial distress score (HSCL-25) b 0.86 (0.85–0.86) 33 (33–33) Psychosocial distress score (HSCL-25) b −1.548 ± 1.418 (118) 0.277 Wellbeing score (MHC-SF) 2.01 (1.39–2.89) 6 (4–13) Wellbeing score (MHC-SF) 4.707 ± 1.816 (119) 0.011 Self-efficacy score: Self-care sub-score 2.02 (1.22–3.35) 6 (4–25) Self-efficacy score: Self-care sub-score 0.801 ± 0.415 (116) 0.056 Self-efficacy score: Infant care sub-score 1.55 (0.68–3.54) 13 (−11–5) Self-efficacy score: Infant care sub-score −0.128 ± 0.380 (75) 0.737 Self-efficacy score: Total 1.14 (0.84–1.56) 50 (−20–13) Infant stimulation score 0.242 ± 0.395 (74) 0.611 Infant stimulation score 1.2 (0.90–1.60) 33 (− 50–17) aIncreases in the HSCL-25, MHC-SF, self-efficacy and infant stimulation scores indicate greater distress, greater wellbeing, greater self-efficacy, and greater aIncreases in the HSCL-25, MHC-SF, self-efficacy and infant stimulation scores maternal engagement in infant stimulation activities, respectively. b indicate greater distress, greater wellbeing, greater self-efficacy, and greater Depression and anxiety sub-scores with similar findings, namely non- maternal engagement in infant stimulation activities, respectively. b Analyses significant association with attendance; data not shown with HSCL-25 anxiety and depression sub-scores were not significant Chomat et al. BMC Women's Health (2019) 19:53 Page 10 of 15

can do. This is what has most caught my attention. It manage their anger and to not make others feel bad by doesn’t leave my mind. insulting or hitting them.

Barriers and potential untoward effects Neither the Improved emotional health and wellbeing Many mothers nor the leaders were aware of any untoward ef- mothers and leaders thought the sessions helped them fects of participation. gain perspective and agonize less over their worries. One Several women were initially anxious for not knowing woman shared: “I used to feel my head didn’t work, I other participants and worried about sensitive informa- was forgetting everything. When I started attending [the tion being shared outside the group. Kinships within the Circles], my heart stopped hurting from all the sadness same group (i.e. in-laws) sometimes limited sharing and my head stopped hurting from thinking too much, freely. Additionally, tensions existed between some par- “ and I could start thinking clearly again, and do what I ticipants, as one woman described: Maybe she does not ’ needed to do.” One leader mentioned: “Many women like me, because she sells pigs and doesn t want me to ” “ leave their homes heavy with worries, sadness and prob- sell pigs too. Another expressed, At first we felt embar- lems. But with the activities they start laughing and feel- rassed and fearful to speak. For example, one woman al- ing happy.” Another shared: “The sessions help them ways tried to make us feel bad. She made fun of us when ” relax, and when they leave, they go with a smile. You we could not do or say things well. However, these fears “ can see the impact on their faces.” and tensions tended to abate over time; Little by little I started feeling more comfortable with the other women, we shared concerns and enjoyed being together.” Improved self-efficacy Several women mentioned they A few women felt intimidated by lack of Spanish flu- used to spend a lot of time at home thinking about their ency and a lower level of literacy. More literate women problems without knowing what to do. Many found it participated more actively. As mentioned by one leader, useful to hear what others go through and share experi- “at first they felt insecure or embarrassed, and said they “ ’ ences and solutions: It helps to get things off one s couldn’t do things we asked them to do, but now they ’ chest, and speak about ones problems with others, to do them, and they laugh and have fun, as if they lost not make problems bigger than they are, and gain know- fear, and little by little realized they can do many ” ledge that can help solve them. A young leader shared things.” “ that her participation had helped her plan her life and Some women had to overcome family restrictions to ” think about her dreams. participate, mostly from mothers-in-laws. A few partici- pated secretly: Improved social support and relationships Women and leaders mentioned that most women in the commu- There are people who make fun of women for nity have few people to turn to for support. Several participating in such activities, who ask, “Why do you women mentioned having no one, while others only had go there? Don’t you have any work? What good is it their husband or mother. Most thought the intervention for you?” Here in the community people make fun of allowed them to develop trusting relationships with what you do all the time. And so, I am afraid. I won’t other women and to feel listened to and newly sup- lie to you, I haven’t told anyone that I am ported. One woman commented, “It was especially nice participating. sharing with other women. Often, we do not get along well with other women. But here, through the activities, Interestingly, most women were able to obtain their you start getting along with other women and you get to partners’ consent. One woman stated, “When I asked for know them. This doesn’t happen often here.” my husband’s authorization, he asked me what I was go- A leader summarized, “The women realized that it was ing to do there, and I explained everything to him, and worth it to come to the sessions and share advice with he told me that it was good, that I was going to be learn- others, and little by little more and more give advice to ing things, that they were very good things to be learn- one another and share experiences and help one an- ing about.” Several husbands reminded the women to go other.” Another mentioned, “The women accepted one the sessions and supported them, so they could leave another as sisters. There were some who were a bit home for a while. proud at first, and others who were shy, but together Sharing the sessions’ content with husbands and they all pulled each other up.” mothers-in-law sometimes increased their support. Several women mentioned that the Circles helped Inclement weather and harvesting presented occa- them improve their relationship with their children and sional barriers to participation. Although poor attend- other family members, because they learned to better ance was usually explained by time constraints and Chomat et al. BMC Women's Health (2019) 19:53 Page 11 of 15

women’s workload, not offering gifts or material goods community’s trust and support, allowing them to access made it harder to motivate mothers. Not having help mothers and take on their new role with relative ease with childcare was also a barrier. and increasing the intervention’s cultural safety and ac- ceptability. The impact that participating in the interven- Discussion tion (first as participants and then as leaders) had on the Principal findings circle leaders’ own wellbeing validates using a cascade The intervention was feasible, acceptable and potentially approach for its delivery and speaks to the need for also efficacious in that it was reported as providing a unique addressing community-based health professionals’ psy- environment for leaders and women participants to chosocial health needs. learn, increase their self-esteem, improve their emotional To the best of our knowledge, this is the first report of wellbeing, discuss and solve their problems, and gain paired group leadership of a group psychosocial inter- new sources of support and friendship. The intervention vention aiming to represent both formal and informal seems to have also increased maternal wellbeing, health systems, and of traditional midwives (comadro- self-efficacy and engagement in early infant stimulation nas) acting as delivery-agents. The systemic neglect of activities; it also had a clustered reduction on psycho- the role of traditional culture in health has been de- social distress. scribed as the single biggest barrier to advancement of the highest attainable standard of health worldwide, es- Community participation pecially among marginalized groups [51]. In Guatemala, This pilot demonstrated acceptability and feasibility of relations between formal and traditional providers are intervention co-design by local women, in a historically often tense due to differing approaches to health, a long excluded population rarely consulted in decision-making history of discrimination and devaluation of indigenous processes relating to their own health. The fields of so- knowledge and practices [20]. The overwhelming re- cial community psychology, critical psychology and lib- course to comadronas by indigenous women testifies to eration psychology have long argued for psychosocial local cultural norms and preferences and greater trust in interventions that contribute to strengthening people’s traditional practices [52–54]. Comadronas’ unique con- possibilities for solving their own problems (conscienti- tributions to women’s psychosocial health would be zation, problematization) in contexts of oppression, worth elucidating further, as would be their ability to struggle and post-colonialism [40, 41] and allow them to transmit resilience factors and endogenous resources become active agents in their own transformation (social rooted in the local context. mobilization) [42]. This is of particular relevance to Guatemala’s indigenous populations, where most psy- Multi-modal collective approach chosocial problems can be traced to the daily stresses of Our pilot study suggests that a multi-modal approach is poverty, discrimination, structural violence and a weak- acceptable, feasible and effective. The small number of ened post-conflict social fabric [43, 44]. The circle psychological interventions in LMICs – with none in- leaders assumed a role as catalysts of change. Interven- cluding indigenous or other marginalized populations of tions that address psychosocial determinants of health Latin America – limits their generalizability to our and wellbeing (i.e. lack of social support, or poor population. A meta-analysis combining trials from self-esteem, self-efficacy and problem-solving skills) are high-income countries (UK, Australia, Canada, USA, likely to have a long-term impact on the prevalence of Germany) and two LMICs (India [28], China [55]) sug- perinatal common mental health disorders and on ma- gests that individual, multi-contact, and interpersonal ternal and child health [45, 46]. therapy-based interventions may be most effective in preventing postnatal depression [56]. A recent Lay health workers as circle leaders meta-analysis of psychological interventions delivered by Our findings add to the accruing evidence from LMIC non-specialist mental health care providers in LMICs that non-mental health specialists such as CHWs [30– found a pooled reduction in maternal depression, but 32] and local women peers [22, 28, 47, 48] can be effect- the heterogeneity of approaches did not permit compari- ive delivery-agents of psychosocial interventions, includ- sons between modalities [5]. Interventions based on bio- ing group interventions [28, 48–50]. This has important medical models of mental illness have proven implications in yet another context where health profes- insufficient for addressing the needs of indigenous com- sionals are scarce [15] and where populations are add- munities [57, 58], and there have been calls instead for a itionally weary of consulting formal health services [20]. collective, holistic, strengths-based approaches rooted in As in other studies [22, 26, 31], our leaders received fo- cultural identity [59–61]. cused training and ongoing supervision. They shared Our own approach allowed local women to find in the mothers’ sociocultural context and already held their Women’s Circles a space that responded to their Chomat et al. BMC Women's Health (2019) 19:53 Page 12 of 15

individual needs and interests, within a mutually sup- community-based maternal health programs [63] and portive environment. A group rather than relying on CHWs as delivery agents could reduce costs individual-focused intervention emerged as the delivery and ease referral to specialist care [27, 64], but also run method of choice in our setting, contributing to the ac- the risk of overburdening fragile health systems, espe- cruing evidence from LMICs that group interventions cially as psychosocial interventions are human-resource can be effective in improving maternal mental health intensive. Intervention co-design may be challenging to [23]. Popular education and arts-based methods reproduce in an institutional setting. Mechanisms to ef- emerged as particularly powerful tools to facilitate fectively support circle leaders to deliver the intervention women’s engagement with the stresses present in their within their communities need to be further explored. daily lives and explorations of a better future. Women’s interest in having more productive activities included in the intervention could be explored as a Strengths and limitations self-sustaining income-generating mechanism. The screening tool may not have detected all truly eli- Finally, the pilot was conducted within a specific context gible women and included only those available to join and we need to use caution in generalizing findings to the Circles. We may have excluded working mothers, other settings. The intervention will need to be adapted to women facing particularly harsh living conditions, the diverse contexts of Guatemala to enable scaling-up. It women not given permission to participate, or women would also benefit from complementary enabling strat- with poor levels of trust. Session attendance was not egies; psychological interventions alone may not be suffi- ideal; better selection of women based on interest and cient when major contributing factors to women’s need may help increase retention, as might be adding in psychosocial distress are systemic and structural [10]. more productive activities, as suggested by participants. Where strong gender inequalities exist, it may be unrealis- Local acceptability of the intervention was likely influ- tic to expect an intervention to empower women in a way enced by human resource elements that may be hard to that they are individually able to negotiate for a change in replicate, built by project lead over the course of many their lives [23]; involving men and communities is critical. years. The pilot trial was not powered to test statistically sig- Conclusions nificant differences in outcome variables, but rather the To the best of our knowledge, this is the first psycho- intervention’s feasibility and acceptability – making us social intervention that engaged end-users as partners in wary of over-interpreting the measured impact. We program co-design, helping to guarantee cultural safety recognize the limits of our statistical approach (i.e. small and acceptability. An important innovation in mental power, lack of clinical corollary, assumptions of linear health, the approach has special relevance in settings regression), however the consistency of findings across without formal mental health services. The Women’s methods used (primary, secondary and supplementary Circle intervention emerges as a promising strategy in analyses) as well as with qualitative finding strengthens end-user engagement and community-based mental their credibility. health promotion and prevention, The HSCL-25 instrument may have lacked sensitivity This study illustrates the feasibility of a holistic, in our patient population, something that we will need community-based, peer-led psychosocial intervention for to investigate further. We do not provide comparisons indigenous women in Latin America. Research findings with other studies because we consider that sociocul- with Maya mothers in Guatemala suggest that women’s tural contexts would be so different that it makes this groups can be leveraged as a critical space where exercise redundant and impossible to reach conclusions. mothers can engage in concrete actions to transform The main dynamic behind impact may have been their lives. If, as postulated, high levels of psychosocial women’s empowerment – a challenging construct to adversity affect not only a mother’s wellbeing but also quantify. We considered the possibility of positive soci- her infant’s growth, development and life opportunity, ability [22], with participants reporting more positive increasing her ability to overcome adversity and psycho- outcomes so that the project might continue. Circle social distress opens new possibilities for breaking vi- leaders’ own aspirations for continued employment may cious cycles of poverty, illness, and mental distress that have resulted in them painting a more positive picture of plague marginalized communities. their experiences. Although the use of non-mental health specialists is a Additional files potentially low-cost strategy to increase women’s access to evidence-based psychosocial care, its sustainability Additional file 1: Table S1. Themes and content overview of Women’s and scalability will need to be further explored [62]. Circles. This table outlines the contents (themes and objectives) of the 10 ’ Strategic nesting of the intervention into existing Women s Circle sessions. (DOCX 17 kb) Chomat et al. BMC Women's Health (2019) 19:53 Page 13 of 15

Additional file 2: Table S2. Linear Regression Assumptions: Absence of Consent for publication multicollinearity (variance inflation factor, or VIF < 2.5), Independence of Not applicable. residuals (Durbin-Watson statistic between 1 and 3), Variance of residuals, or homoscedasticity (scatterplot of residuals), and Normal distribution of Competing interests residuals (normal P-P plot of residuals). A. Linear regression assumptions The authors declare that they have no competing interests. for Table 4: Multiple linear regression models, adjusted for maternal age, area of residence and baseline score. B. Linear regression assumptions for Publisher’sNote Table 5: Multiple linear regression models, adjusted for maternal age, area Springer Nature remains neutral with regard to jurisdictional claims in of residence and baseline score. This table presents all assumptions that published maps and institutional affiliations. were tested prior to carrying out the multiple linear regression analyses, ’ presented in the manuscript s Tables 4 and 5. These include, as described Author details in the manuscript under Analyses and in the Additional file Table: 1Participatory Research at McGill (PRAM), Department of Family Medicine, Absence of multicollinearity (variance inflation factor, or VIF < 2.5), McGill University, 5858 Chemin de la Côte-des-Neiges-3rd floor, Suite 300, Independence of residuals (Durbin-Watson statistic between 1 and 3), Montréal, QC H3S 1Z1, Canada. 2CIET International Guatemala, 5ª calle 14-35, Variance of residuals, or homoscedasticity (scatterplot of residuals), and apartamento 304, Edificio Las Tapias, zona 3, Quetzaltenango, Guatemala. Normal distribution of residuals (normal P-P plot of residuals). 3Centro de Investigación de Enfermedades Tropicales (CIET), Universidad (DOCX 27 kb) Autónoma de Guerrero, Acapulco, Mexico. 4Research Center for the Prevention of Chronic Diseases (CIIPEC), Institute of Nutrition of Central America and Panama (INCAP), Calzada Roosevelt 6-25 zona 11, Apartado Abbreviations Postal 1188, Guatemala City, Guatemala. 5Department of Psychiatry and CBT: Cognitive behavioural therapy; CHW: Community health workers; HSCL- Division of Social and Transcultural Psychiatry, McGill University, Montréal, 25: Hopkins Symptom Checklist-25; INCAP: Institute of Nutrition of Central Canada. 6Centre for Global Mental Health, King’s College London, De America and Panama; IPT: Interpersonal therapy; LMIC: Low- and middle- Crespigny Park, London SE5 8AF, UK. income countries; MHC-SF: Mental Health Continuum Short Form; UK: United ’ Kingdom; UNICEF: United Nations Children s Fund; USA: United States of Received: 19 December 2018 Accepted: 14 March 2019 America

Acknowledgements References The authors would especially like to thank the 16 circle leaders who were 1. Surkan PJ, Kennedy CE, Hurley KM, Black MM. Maternal depression and early instrumental in co-designing the study, testing all methodologies as Women childhood growth in developing countries: systematic review and meta- Circle participants, and piloting the intervention in their communities. The analysis. Bull World Health Organ. 2011;89(8):608–15. authors would also like to thank study participants and their families; in- 2. Parsons CE, Young KS, Rochat TJ, Kringelbach ML, Stein A. Postnatal volved communities; Dr. Victor Lopez for his invaluable inputs to our original depression and its effects on child development: a review of evidence from protocol; the GCC Global Mental Health and Saving Brains communities; low- and middle-income countries. Br Med Bull. 2012;101:57–79. Marta Escobar, Alejandra Maldonado, Maria Maldonado Garcia and the stu- 3. Fox M, Entringer S, Buss C, DeHaene J, Wadhwa PD. Intergenerational dents who helped in the field; and Mr. Robert Ledogar, Ms. Kara Anderson transmission of the effects of acculturation on health in Hispanic Americans: and Drs. Luis Paiz Bekker and Czarina Thelen for critically reviewing the a fetal programming perspective. Am J Public Health. 2015;105(Suppl 3): manuscript. S409–23. 4. Stein A, Pearson RM, Goodman SH, Rapa E, Rahman A, McCallum M, Funding Howard LM, Pariante CM. Effects of perinatal mental disorders on the fetus This study was supported by a Grand Challenges Canada Global Mental and child. Lancet. 2014;384(9956):1800–19. Health seed grant (grant number 0333–04). The funding body had no role in 5. Rahman A, Fisher J, Bower P, Luchters S, Tran T, Yasamy MT, Saxena S, the design of the study, in data collection, analysis or interpretation, or in Waheed W. Interventions for common perinatal mental disorders in women writing the manuscript. in low- and middle-income countries: a systematic review and meta- analysis. Bull World Health Organ. 2013;91(8):593–601I. 6. Pearson RM, Evans J, Kounali D, Lewis G, Heron J, Ramchandani PG, Availability of data and materials O'Connor TG, Stein A. Maternal depression during pregnancy and the The datasets used during the current study are available from the postnatal period: risks and possible mechanisms for offspring depression at corresponding author on reasonable request. age 18 years. JAMA Psychiatry. 2013;70(12):1312–9. 7. Fleuriet KJ, Sunil TS. Perceived social stress, pregnancy-related anxiety, ’ Authors contributions depression and subjective social status among pregnant Mexican and AMC co-designed the study, led all training and research activities, analyzed Mexican American women in South Texas. J Health Care Poor Underserved. and interpreted the data and drafted the manuscript. AIM co-designed and 2014;25(2):546–61. co-implemented all research activities, entered the data, and provided critical 8. Dunkel Schetter C. Psychological science on pregnancy: stress processes, insights in data interpretation. MRZ co-designed the study, participated in biopsychosocial models, and emerging research issues. Annu Rev Psychol. discussions on the interpretation of data, and revised the manuscript. DP co- 2011;62:531–58. designed the study, participated in early discussions on the interpretation of 9. Stewart RC. Maternal depression and infant growth: a review of recent data, prior to his passing away on January 27, 2016. PL assisted in study de- evidence. Matern Child Nutr. 2007;3(2):94–107. sign and in the analysis and interpretation of the qualitative data. AB assisted 10. Fisher J, Cabral de Mello M, Patel V, Rahman A, Tran T, Holton S, Holmes W. in the analysis and interpretation of the qualitative data and in drafting the Prevalence and determinants of common perinatal mental disorders in manuscript. NA assisted in quantitative analyses and data interpretation and women in low- and lower-middle-income countries: a systematic review. in drafting the manuscript. RA co-designed the study, advised in the method Bull World Health Organ. 2012;90(2):139G–49G. of analysis, and revised the manuscript. All authors read and approved the 11. Chmielowska M, Fuhr DC. Intimate partner violence and mental ill health final manuscript. among global populations of indigenous women: a systematic review. Soc Psychiatry Psychiatr Epidemiol. 2017;52(6):689–704. Ethics approval and consent to participate 12. Nierop A, Wirtz PH, Bratsikas A, Zimmermann R, Ehlert U. Stress-buffering Ethical approval was obtained from the Douglas Mental Health Institute effects of psychosocial resources on physiological and psychological stress (McGill University affiliate) and the Guatemala-based Institute of Nutrition of response in pregnant women. Biol Psychol. 2008;78(3):261–8. Central America and Panama (INCAP). Informed written consent was received 13. Campos B, Schetter CD, Abdou CM, Hobel CJ, Glynn LM, Sandman CA. from all participants to participate, using consent forms approved by these Familialism, social support, and stress: positive implications for pregnant research ethics boards. Latinas. Cultur Divers Ethnic Minor Psychol. 2008;14(2):155–62. Chomat et al. BMC Women's Health (2019) 19:53 Page 14 of 15

14. MSPAS. Encuesta Nacional de Salud Materno Infantil (ESNMI) 2014–2015. 35. Derogatis LR, Lipmann RS, Rickels K. The Hopkins symptoms checklist Guatemala: Ministerio de Salud Pública y Asistencia Social, Instituto Nacional (HSCL): a self-report inventory. Behav Sci. 1974;19:1–15. de Estadística, Secretaría de Planificación y Programación de la Presidencia; 36. Keyes CLM. Atlanta: Brief description of the mental health continuum short 2016. form (MHC-SF), 2009. Available: http://www.sociology.emory.edu/ckeyes/. 15. WB. Guatemala Poverty Assessment. Report No.43920-GT. Central America [On–line, retrieved 20 March 2019]. Department, Poverty Reduction and Economic Manager, Latin America and 37. UNICEF: Multiple indicator cluster surveys (MICS). 2017. the Caribbean Region. World Bank, March 18, 2009. Available at: http:// 38. logit.gee: Generalized Estimating Equation for Logistic Regression. http://ftp. siteresources.worldbank.org/INTLACREGTOPPOVANA/Resources/ uni-bayreuth.de/math/statlib/R/CRAN/doc/vignettes/Zelig/logit.gee.pdf. GuatemalaPovertyAssessmentEnglish.pdf. Accessed 20 Mar 2019. Accessed 20 Mar 2019. 16. Lopez V, Calgua E, Garcia C. Salud Mental en Guatemala (Resumen Ejecutivo 39. Charmaz K. Constructing grounded theory : a practical guide through de la Encuesta Nacional de Salud Mental), Documento para la Docencia, qualitative analysis. London; Thousand Oaks: Sage Publications; 2006. Facultad de Ciencias Médicas, USAC. Guatemala, October 2009. 40. Freire P. Pedagogy of the oppressed, new rev. 20th-anniversary edn. New 17. Chomat AM. Maternal stressors impact maternal wellbeing and cortisol and York: Continuum; 1993. infant growth in rural Guatemala: insights from qualitative and quantitative 41. Martín-Baró I. Writings for a liberation psychology. New York: Harvard approaches. Montréal: McGill University; 2016. University Press; 1996. 18. WHO. Mental health atlas: Guatemala. Guatemala: World Health 42. Burton M, Kagan C. Methods for liberation: critical consciousness in action. Organization, Department of Mental Health and Substance Abuse; 2017. In: Montero M, Sonn CC, editors. Psychology of liberation. New York: 19. PAHO, WHO. Informe sobre el sistema de salud mental en Guatemala 2011. Springer-Verlag; 2009. p. 51–72. Guatemala: Organización Panamericana de la Salud (OPS), Organización 43. Anckermann S, Dominguez M, Soto N, Kjaerulf F, Berliner P, Mikkelsen EN. Mundial de la Salud (OMS); 2011. Psychosocial support to large numbers of traumatized people in post- 20. Cabrera LO, Rosales P, Echeverría G, Palma S, López S, Haeussler R, Calan M. conflict societies: an approach to community development in Guatemala. J Normas con Pertinencia Cultural … hacia la interculturalidad. In. Edited by Community App Soc Psychol. 2005;15:136–52. Ministerio de Salud Publica y Asistencia Social DdRdlPdAalP; 2009. 44. Sullivan ER. Sadness in the highlands: a study of depression in Nueva Santa 21. MMN. Claves para la transformación de los sistemas de salud en América Catarina Ixtahuacan. In: Adams WR, Hawkins JP, editors. Health care in Maya Latina. Bolivia, Guatemala y Perú: tres experiencias, una sola acción integral Guatemala: confronting medical pluralism in a developing country. Norman: e incluyente en atención primaria de salud, vol. 2017. Guatemala: Medicus University of Oklahoma Press; 2007. Mundi Navarra; 2013. 45. Prost A, Colbourn T, Seward N, Azad K, Coomarasamy A, Copas A, 22. Atif N, Krishna RN, Sikander S, Lazarus A, Nisar A, Ahmad I, Raman R, Fuhr Houweling TA, Fottrell E, Kuddus A, Lewycka S, et al. Women’s groups DC, Patel V, Rahman A. Mother-to-mother therapy in India and Pakistan: practising participatory learning and action to improve maternal and adaptation and feasibility evaluation of the peer-delivered thinking healthy newborn health in low-resource settings: a systematic review and meta- Programme. BMC Psychiatry. 2017;17(1):79. analysis. Lancet. 2013;381(9879):1736–46. 23. Clarke K, King M, Prost A. Psychosocial interventions for perinatal common 46. Seward N, Neuman M, Colbourn T, Osrin D, Lewycka S, Azad K, Costello A, mental disorders delivered by providers who are not mental health Das S, Fottrell E, Kuddus A, et al. Effects of women’s groups practising specialists in low- and middle-income countries: a systematic review and participatory learning and action on preventive and care-seeking meta-analysis. PLoS Med. 2013;10(10):e1001541. behaviours to reduce neonatal mortality: a meta-analysis of cluster- 24. van Ginneken N, Tharyan P, Lewin S, Rao GN, Meera SM, Pian J, randomised trials. PLoS Med. 2017;14(12):e1002467. Chandrashekar S, Patel V. Non-specialist health worker interventions for the 47. Ali NS, Ali BS, Azam IS, Khuwaja AK. Effectiveness of counseling for anxiety and care of mental, neurological and substance-abuse disorders in low- and depression in mothers of children ages 0-30 months by community workers in middle-income countries. Cochrane Database Syst Rev. 2013;11:CD009149. Karachi, Pakistan: a quasi experimental study. BMC Psychiatry. 2010;10:57. 25. Mao HJ, Li HJ, Chiu H, Chan WC, Chen SL. Effectiveness of antenatal 48. Futterman D, Shea J, Besser M, Stafford S, Desmond K, Comulada WS, Greco emotional self-management training program in prevention of postnatal E. Mamekhaya: a pilot study combining a cognitive-behavioral intervention depression in Chinese women. Perspect Psychiatr Care. 2012;48(4):218–24. and mentor mothers with PMTCT services in South Africa. AIDS Care. 2010; 26. Rahman A, Malik A, Sikander S, Roberts C, Creed F. Cognitive behaviour 22(9):1093–100. therapy-based intervention by community health workers for mothers with 49. Chibanda D, Shetty AK, Tshimanga M, Woelk G, Stranix-Chibanda L, depression and their infants in rural Pakistan: a cluster-randomised Rusakaniko S. Group problem-solving therapy for postnatal depression controlled trial. Lancet. 2008;372(9642):902–9. among HIV-positive and HIV-negative mothers in Zimbabwe. J Int Assoc 27. Rojas G, Fritsch R, Solis J, Jadresic E, Castillo C, Gonzalez M, Guajardo V, Provid AIDS Care. 2014;13(4):335–41. Lewis G, Peters TJ, Araya R. Treatment of postnatal depression in low- 50. Morris J, Jones L, Berrino A, Jordans MJ, Okema L, Crow C. Does combining income mothers in primary-care clinics in Santiago, Chile: a randomised infant stimulation with emergency feeding improve psychosocial outcomes controlled trial. Lancet. 2007;370(9599):1629–37. for displaced mothers and babies? A controlled evaluation from northern 28. Tripathy P, Nair N, Barnett S, Mahapatra R, Borghi J, Rath S, Rath S, Gope R, Uganda. Am J Orthop. 2012;82(3):349–57. Mahto D, Sinha R, et al. Effect of a participatory intervention with women's 51. Napier AD, Ancarno C, Butler B, Calabrese J, Chater A, Chatterjee H, Guesnet groups on birth outcomes and maternal depression in Jharkhand and Orissa, F, Horne R, Jacyna S, Jadhav S, et al. Culture and health. Lancet. 2014; India: a cluster-randomised controlled trial. Lancet. 2010;375(9721):1182–92. 384(9954):1607–39. 29. Lara MA, Navarro C, Navarrete L. Outcome results of a psycho-educational 52. Chomat AM, Solomons NW, Crowley C, Montenegro G, Bermudez O. Maternal intervention in pregnancy to prevent PPD: a randomized control trial. J health and health-seeking behavior among indigenous mam mothers from Affect Disord. 2010;122(1–2):109–17. Quetzaltenango, Guatemala. Rev Panam Salud Publica. 2014;35(2):113–20. 30. Baker-Henningham H, Powell C, Walker S, Grantham-McGregor S. The effect 53. Glei DA, Goldman N. Understanding ethnic variation in pregnancy-related of early stimulation on maternal depression: a cluster randomised care in rural Guatemala. Ethn Health. 2000;5(1):5–22. controlled trial. Arch Dis Child. 2005;90(12):1230–4. 54. Perén HI. Revival of Maya medicine and impact for its social and political 31. Cooper PJ, Tomlinson M, Swartz L, Landman M, Molteno C, Stein A, recognition (in Guatemala). A case study commissioned by the Health McPherson K, Murray L. Improving quality of mother-infant relationship and Systems Knowledge Network; 2007. infant attachment in socioeconomically deprived community in South 55. Tam WH, Lee DTS, Chiu HFK, Ma KC, Lee A, Chung TKH. A randomised controlled Africa: randomised controlled trial. BMJ. 2009;338:b974. trial of educational counselling on the management of women who have 32. Rahman A, Iqbal Z, Roberts C, Husain N. Cluster randomized trial of a suffered suboptimal outcomes in pregnancy. BJOG. 2003;110(9):853–9. parent-based intervention to support early development of children in a 56. Dennis CL, Dowswell T. Psychosocial and psychological interventions for low-income country. Child Care Health Dev. 2009;35(1):56–62. preventing postpartum depression. Cochrane Database Syst Rev. 2013;2:CD001134. 33. CIHR. CIHR guidelines for health research involving aboriginal people. 57. Calabrese JD. Clinical paradigm clashes: ethnocentric and political barriers Ottawa: Canadian Institute of Health Research; 2007. to native American efforts at self-healing. Ethos. 2008;36(3):334–53. 34. Minkler M, Wallerstein N. Community-based participatory research for 58. Gone JP. Dialogue 2008 introduction: mental health discourse as Western health: from process to outcomes. 2nd ed. San Francisco: Jossey-Bass; 2008. cultural proselytization. Ethos. 2008;36(3):310–5. Chomat et al. BMC Women's Health (2019) 19:53 Page 15 of 15

59. Bourke L, Humphreys JS, Wakerman J, Taylor J. From ‘problem-describing’ to ‘problem-solving’: challenging the ‘deficit’ view of remote and rural health. Aust J Rural Health. 2010;18(5):205–9. 60. Kirmayer LJ, Sehdev M, Isaac C. Community resilience: models, metaphors and measures. Int J Indig Health. 2009;5(1):62. 61. Pyett P, Waples-Crowe P, van der Sterren A. Challenging our own practices in indigenous health promotion and research. Health Promot J Austr. 2008; 19(3):179–83. 62. Chowdhary N, Sikander S, Atif N, Singh N, Ahmad I, Fuhr DC, Rahman A, Patel V. The content and delivery of psychological interventions for perinatal depression by non-specialist health workers in low and middle income countries: a systematic review. Best Pract Res Clin Obstet Gynaecol. 2014;28(1):113–33. 63. Hanlon C. Maternal depression in low- and middle-income countries. Int Health. 2013;5(1):4–5. 64. Patel V, Weiss HA, Chowdhary N, Naik S, Pednekar S, Chatterjee S, De Silva MJ, Bhat B, Araya R, King M, et al. Effectiveness of an intervention led by lay health counsellors for depressive and anxiety disorders in primary care in Goa, India (MANAS): a cluster randomised controlled trial. Lancet. 2010; 376(9758):2086–95.